• ภาษาไทย
  • English

Geo-Informatics and Space Technology Development Agency (Public Organization):GISTDA

GISTDA To be Value based Organization in Geo-space Intelligence Deliveries by 2020

Field Serve (Part 4) Segmentation of the interested “Rice Plot” from the image recorded by Field Server

Field Serve (Part 4) Segmentation of the interested “Rice Plot” from the image recorded by Field Server

In the previous part, it indicates the calculation method of Vegetation Phenology gained from rice plot, which made us realized that Vegetation Index used in finding the Green level of each image is Excessive Green (ExG). It could calculate Visible wavelength or RGB. As for the process we use in calculation Vegetation Index of each image from Field Server. In order to obtain an effective and usable value, we need to choose only rice plot (divide the part that is not rice plot). As shown in the picture number 1.

 

รูปที่ 1 รูปภาพที่บันทึกได้จาก Field Server ประกอบด้วย บริเวณ “แปลงข้าว” และ “ไม่ใช่แปลงข้าว”

รูปที่ 1 รูปภาพที่บันทึกได้จาก Field Server ประกอบด้วย บริเวณ “แปลงข้าว” และ “ไม่ใช่แปลงข้าว”

This part explains about fundamental method used in segmenting AOI: Area of Interest. From the image which is “Rice Plot.” We probably recognize from the last part that each image pixel could be calculated into ExG Vegetation Index, showing in picture no.2 (Left). We could see that area of image pixel that is rice plot will have ExG value (approximately) more than 0.2 when considering from rice plot in growing status (plenty of green color). When selecting group of image pixel which has ExG value more than 0.2 by specifying the part that is not in the described condition as black color, the result is as showed in the second picture (Right).

 

รูปที่ 2 รูปภาพ Field Server ถูกคำนวณและแทนด้วยค่า ExG สำหรับแต่ละจุดภาพ (ซ้าย)

รูปที่ 2 รูปภาพ Field Server ถูกคำนวณและแทนด้วยค่า ExG สำหรับแต่ละจุดภาพ (ซ้าย) เมื่อพิจารณาค่า ExG มากกว่า 0.2 จะได้กลุ่มจุดภาพบริเวณแปลงข้าว (ขวา)

The result will show a group of majority image pixel of rice plot which is AOI. If there is need in increasing the correctness of choosing area, it could be done by evaluating fundamental image using dilation and erosion, which reducing small pixel of noise or also known as Pepper noise. Image no.3 shows the result of image evaluation in order to reduce the described noise. In later process, the area of the largest image pixel will be chosen to specify rice plot area while the other group of image pixel would be abandon.

รูปที่ 3 ผลลัพธ์ของการลดสัญญาณรบกวน และการเลือกบริเวณที่ใหญ่ที่สุดเป็น “บริเวณสนใจ” (แปลงข้าว)

รูปที่ 3 ผลลัพธ์ของการลดสัญญาณรบกวน และการเลือกบริเวณที่ใหญ่ที่สุดเป็น “บริเวณสนใจ” (แปลงข้าว)

In the next part, we will follow up on how Vegetation Phenology graph could be used in evaluation for the purpose of making cropping calendar of rice, knowing the beginning and ending date of cropping cycle. Until next time!

Note
• The process of dividing rice plot is the first step before each image would be calculated as “Average Vegetation Index” to create “Vegetation Phenology Graph” in the next step.
• AOI of rice plot will be specify by Binary mask in order to know which image pixel “is” or “isn’t” rice plot. Then all image pixel of rice plot will be calculated to Vegetation Index and use to find the average of “Average Vegetation Index” for each image.
• Using “Average Vegetation Index” of the whole rice plot causing durability to noise better than using the chosen group of image pixel.
• Segmentation rice plot area as described earlier is considered as easily fundamental method. However, nowadays there are a lot of more effective and accurate methods. People who are interested in this topic could research further using keyword “Image segmentation”

Tags: